Point 1: I don't understand the logic. You want square geometry so you want to set vertical overscan to equal horizontal overscan.
Also on a real CRT, the yoke isn't perfectly and, especially the top area on a Sony, often times has misconvergance so having lines of colors is not useful. The old monoscope, white boarder along the edge of the raster, was very useful for both geometry like H-TRP and convergance (TLH) and even convergance strips, better than even the dedicated convergance patterns.
Point 7: from ChatGPT
"A human usually detects a distorted circle (oval/ellipse) faster than noticing that a square is “just a rectangle”, but those are slightly different kinds of judgment.
- Circle → oval: this is mostly perceptual. Your visual system is very sensitive to symmetry. Even a small stretch makes a circle look “off” immediately.
- Square → rectangle: this is mostly categorical. A square is a rectangle by definition (a special case), so the issue is not seeing it — it’s whether your brain is thinking in the broad category or the narrower one."
The squares are more difficult to use and measuring tape is slow.